Synonymous codon substitutions modulate transcription and translation of a divergent upstream gene by modulating antisense RNA production

Abstract

Synonymous codons were originally viewed as interchangeable, with no phenotypic consequences. However, substantial evidence has now demonstrated that synonymous substitutions can perturb a variety of gene expression and protein homeostasis mechanisms, including translational efficiency, translational fidelity, and cotranslational folding of the encoded protein. To date, most studies of synonymous codon-derived perturbations have focused on effects within a single gene. Here, we show that synonymous codon substitutions made far within the coding sequence of Escherichia coli plasmid-encoded chloramphenicol acetyltransferase (cat) can significantly increase expression of the divergent upstream tetracycline resistance gene, tetR. In four out of nine synonymously recoded cat sequences tested, expression of the upstream tetR gene was significantly elevated due to transcription of a long antisense RNA (asRNA) originating from a transcription start site within cat. Surprisingly, transcription of this asRNA readily bypassed the native tet transcriptional repression mechanism. Even more surprisingly, accumulation of the TetR protein correlated with the level of asRNA, rather than total tetR RNA. These effects of synonymous codon substitutions on transcription and translation of a neighboring gene suggest that synonymous codon usage in bacteria may be under selection to both preserve the amino acid sequence of the encoded gene and avoid DNA sequence elements that can significantly perturb expression of neighboring genes. Avoiding such sequences may be especially important in plasmids and prokaryotic genomes, where genes and regulatory elements are often densely packed. Similar considerations may apply to the design of genetic circuits for synthetic biology applications.

Keywords: antisense RNA; gene expression; overlapping divergent promoters; protein homeostasis; synonymous codon substitutions.

Fig. 1.

Synonymous codon substitutions can lead to severe defects in cat gene expression. (A) Representative western blot to detect synonymous CAT mutant protein abundance in whole cell lysates, for 10 synonymous cat sequences. See text for details on sequence design. EV: empty vector with cat coding sequence removed. GAPDH: loading control. Top: Blot image used for quantification in (B). Bottom: Blot image shown with increased contrast, for display purposes only. (B) For each blot, quantification of CAT protein abundance was normalized to GAPDH loading control and reported relative to WT (red bar); see Materials and Methods for details. (C) Correlation of CAT protein accumulation measured in (B) with the mean fold change of cat mRNA level relative to WT, as determined by RT-qPCR. Significance determined by two-tailed Pearson r correlation with 95% CI. (D) There is no significant difference between average plasmid abundance for WT versus selected cat synonymous mutants, as measured by qPCR. For all categorical scatter plots, bar height indicates the median value of at least three biological replicates (*P < 0.05; **P < 0.01; Kruskal–Wallis, uncorrected Dunn’s test). Mutants without a label or labeled “ns” were not significantly different from WT (P > 0.05).

Fig. 2.

Synonymous codon substitutions in cat can increase TetR protein accumulation, which represses cat expression. (A) Schematic of the Tet ON/OFF promoter system used to express cat. Transcription of both tetR and cat is simultaneously repressed by binding of TetR protein (purple) to the two operator sites (green). Yellow vertical lines indicate location of synonymous codon substitutions in the cat coding sequence, far from the canonical promoter region. (B) Representative western blot using an anti-TetR primary antibody to detect TetR protein in whole cell lysates; CAT was detected using an anti-His antibody. Top: Blot image used for quantification in (C). Bottom: Blot image shown with increased contrast, for display purposes only. EV: empty vector with cat coding sequence removed but tetR gene still present. (C) Quantification of TetR protein abundance for each construct, normalized to loading control (GAPDH), relative to the abundance of TetR observed for the WT cat construct on each blot (red bar). For all categorical scatter plots, bar height indicates the median value of at least three biological replicates (*P < 0.05; **P < 0.01; Kruskal–Wallis, uncorrected Dunn’s test). Mutants without a label or labeled ns were not significantly different from WT (P > 0.05). (D) TetR protein level is inversely related to cat mRNA level. Significance determined by two-tailed Pearson r correlation with 95% CI.

Fig. 3.

TetR protein upregulation due to cat synonymous codon substitutions is independent of cat translation and induction but sensitive to plasmid location of the recoded cat sequence. (A) Representative western blot of TetR protein when the cat RBS is nonfunctional. Top: Blot image used for quantification in (B). Bottom: Blot image shown with increased contrast, for display purposes only. (B) Quantification of TetR protein abundance when the cat RBS is nonfunctional ((−) RBS). (C) Western blot of TetR protein in the absence of inducer. Top: Blot image used for quantification in (D). Bottom: Blot image shown with increased contrast, for display purposes only. (D) Quantification of TetR protein in the absence of inducer. (E) Representative western blot of TetR protein abundance when the indicated synonymous cat coding region is cloned into the plasmid backbone of WT-cat. (F) Quantification of TetR protein abundance when the indicated synonymous cat coding region is cloned into the plasmid backbone of WT-cat. TetR abundance was not significantly different from WT (P > 0.05). For all categorical scatter plots, bar height indicates the median value of at least three biological replicates (*P < 0.05; **P < 0.01; Kruskal–Wallis, uncorrected Dunn’s test). Mutants without a label or labeled ns were not significantly different from WT (P > 0.05).

Fig. 4.

Some cat synonymous substitutions significantly enhance tetR transcription. (A) RT-qPCR quantification of intragenic tetR RNA abundance for synonymous mutants relative to WT cat. P_R1 and P_R2: canonical tetR TSSs. P_C: cat TSS. Bold arrows indicate locations of forward and reverse primer; note that reverse primer is designed to bind 41 bp upstream of P_R2. Lengths are not drawn to scale. Note that significant amounts of the intragenic tetR RNA are detected even for the WT cat construct (SI Appendix, Fig. S6). (B) RT-qPCR quantification of total tetR mRNA abundance, relative to WT. As in panel A, bold arrows indicate locations of forward and reverse primers. For all categorical scatter plots, bar height indicates the median value of at least three biological replicates (*P < 0.05; **P < 0.01; Kruskal–Wallis, uncorrected Dunn’s test). Mutants without a label or labeled ns were not significantly different from WT (P > 0.05).

Fig. 5.

Introduction of a transcription terminator in the 5′ region of cat reduces tetR transcription originating from the intragenic TSS and TetR protein abundance. (A) Schematic indicating location of transcription terminator (red). (B) Representative western blot of TetR protein abundance in constructs with or without the transcription terminator. (C) RT-qPCR measurements of the influence of the transcription terminator on the level of tetR mRNA originating from the intragenic TSS, measured using the primer pair shown in Fig. 4A. (D) Quantification of Western blot shown in (B). For all categorical scatter plots, bar height indicates the median value of at least three biological replicates [*P < 0.05; Kruskal–Wallis, uncorrected Dunn’s test; ns = no significant difference (P > 0.05)].

Fig. 6.

Intragenic tetR transcription is affected when the identified core promoter region is altered. (A) Schematic of chimeric cat sequences, each of which exchange the 50 nt region immediately upstream of the identified intragenic TSS. WT, GA, and CAI cat constructs are shown as black, green, and blue, respectively. (B) Quantification of relative abundance of tetR RNA originating from the intragenic TSS, as described in Fig. 4A [*P < 0.05; Kruskal–Wallis, uncorrected Dunn’s test; ns = no significant difference (P > 0.05)].

Fig. 7.

TetR protein abundance corresponds to the level of tetR RNA derived from the intragenic TSS, rather than total tetR RNA (two-tailed Pearson r correlation with 95% CI). (A) No significant correlation was detected between intragenic TSS and total tetR transcript levels. (B) TetR protein abundance is significantly correlated to the level of tetR asRNA arising from the intragenic TSS. (C) TetR protein abundance was not significantly correlated to total tetR mRNA abundance.

Fig. 8.

Model for synonymous substitutions up-regulating transcription and translation of an upstream gene. Synonymous substitutions in the cat coding sequence (orange) can enhance transcription from an internal promoter on the antisense strand. Although this asRNA transcript represents a subset of total tetR mRNA, its level correlates closely with the resulting level of TetR protein. The correlation between asRNA abundance and TetR protein level may occur because (as shown here) the asRNA is translated with higher efficiency than mRNA from the canonical tetR TSS. Alternatively (not shown), the asRNA may act as a noncoding RNA to up-regulate translation of the canonical tetR RNA. Regardless of the precise mechanism, the tetR asRNA provides a mechanism to maintain TetR-mediated transcriptional repression of cat while bypassing repression of tetR.